52 HENRY A. EOWLAND 



giving the tension of the lines of force per square centimetre and square 

 inch of section of the lines. These have been deduced from the formula 



given by Maxwell ' 3 for the tension per square metre, which is 2C 



&~ 



absolute units of force. 

 This becomes 



24655^00000 kil g rammes P er S( l uare centim > I 



} , (12) 



173240000 Ibs. per square inch, 



from which the quantities in the Table were calculated. 



It is seen that the maximum of magnetism of ordinary bar iron is 

 about 175,000 times the unit field, or 177 Ibs. on the square inch, and 

 for nickel 63,000 times, or 22-9 Ibs. on the square inch. For pure iron, 

 however, I think it may reach 180,000, or go even above that. It is 

 seen that one of the Norway rings gave a very high result; this is 

 explained by the following considerations. All the iron rings were 

 welded except this one, which was forged solid from a bar 2 inches 

 wide and then turned. Even the purest bar iron is somewhat fibrous; 

 and between the fibres we often find streaks of scale lying lengthwise 

 in the bar and so diminishing the section somewhat if the ring be 

 welded from the bar; when, however, it is forged solid, these streaks 

 are thoroughly disintegrated; and hence we find a higher maximum 

 of magnetism for a ring of this kind, and one approaching to that of 

 pure iron. But a ring made in this way has to be exposed to so much 

 heating and pounding that the iron is rendered unhomogeneous, and a 

 tail appears to the curve like that in Table III. It is evident that this 

 tail must always show itself whenever the section of the ring is not 

 homogeneous throughout. 



Hence we may conclude that the greatest weight which can be sus- 

 tained by an electromagnet with an infinite current is, for good but not 

 pure iron, 354 Ibs. per square inch of section, and for nickel 46 Ibs. 



Joule 2 * has made many experiments on the maximum sustaining- 

 power of magnets, and has collected the following Table, which I give 

 complete, except that I have replaced the result with his large magnet 

 by one obtained later. 



It is seen that these are all below my estimate, as they should be. 



23 Treatise on Electricity and Magnetism, vol. ii, p. 256. 

 2* Phil. Mag., 1851. 



